These (very rough!) thoughts were raised by reading Onora O'Neill's "A Question of Trust" (see the site bibliography for details and a brief summary of my thoughts), in which she talks about the "crisis of trust" (her inverted commas) in public institutions.
Saying that we "trust someone" is imprecise. "Trusting" isn't enough - you have to trust someone (or an entity) to do something. This can be implicit in a role that entity has taken on, but the implicitness of the relationship assumed by the transaction between the truster and the trustee is dangerous: how can we be certain that all parties have the same expectation?
Trust is about transactions (always?), whether they are repeated or even an agreement not to carry out a particular course of action, and it is the contract around that transaction that carries - holds - that trust. The terms of risk, accountability, ability to roll-back and repudiate the transaction, and track the entities involved, are all significant parts of that contract.
I'm aware that this approach echoes somewhat speech-act theory, and has resonances with Habermas' approach to communication - this is no surprise, given the fact that I've recently been reading his work. I think that we can probably go beyond this, however, and that it may well be useful to examine trust-transactions from the view of communications theory. This is an area I intend to explore in the future.
O'Neill talks about the requirement for individuals to accept "duties" in return for "rights", where one of these rights can be that you should be able to "trust" and institution. I would take this line of thought further and reflect in the contract between truster and trustee the duties of the truster in the contract. Under what circumstances might a trustee be justified in breaking a contract? Who decides this justification, and what recourse is there to accountability in these cases?